Determinants of interaction intention to purchase online in less developed countries: The moderating role of technology infrastructure

Abstract Even though there is significant growth in online purchasing platforms globally, interaction with such platforms is not very popular in less developed countries (LDCs) like Yemen and needs further investigation. Thus, this empirical research aims to investigate the determinants of Yemeni consumers’ interaction intention to purchase online (IIPO) as a context for LDCs. The research developed an extended model of the technology acceptance model (TAM) by adding technology infrastructure (TI), technology anxiety (TA), social influence (SI), and self-efficacy (SE) as independent variables that impact IIPO. The study model further included technology infrastructure as a moderator. The suggested model was investigated through quantitative research with a sample of 273 Yemeni consumers. The results were estimated using partial least squares structural equation modeling (PLS-SEM). The results revealed that TI, SI, and SE have a direct positive impact on IIPO. However, perceived usefulness, ease of use, and TA showed an insignificant direct impact on IIPO. Moreover, technology infrastructure moderates the relationship between (perceived ease of use, TA, SE) and IIPO. A technology infrastructure, in contrast, does not moderate the relationships between other determinants (such as perceived usefulness and SI) and IIPO. The suggested model interpreted 70.5% of the variance in IIPO and 51.7% in perceived usefulness. There is a scarcity of empirical research investigating online purchasing behaviour in LDCs. In Yemen, particularly, the factors influencing Yemeni consumers’ interaction intention to purchase online have not been empirically determined. Moreover, this is one of the first studies to explore the moderate role of technological infrastructure in online purchasing. In this regard, this research claims its uniqueness. More implications were discussed in this study.


Introduction
Globally, there is a significant growth in Internet use through mobile phones and computers. On the part of the firm, the Internet plays a crucial role in enabling firms to be competitive and flexible. It provides a platform for promoting products and gaining knowledge, communication, and decisionmaking quality (Isaac et al., 2018). Internet platforms allow people to freely interact with others and offer multiple ways for marketers to reach and engage with consumers (Appel et al., 2020). The significant growth in Internet platforms led to radical changes in business relationships, particularly in consumers' purchasing habits (Gupta & Arora, 2017;Hanjaya et al., 2019;Isaac et al., 2018;Venkatesh et al., 2022). The consumer's desire to interact with purchasing online via a mobile app or website has increased (Al-Hattami, 2021;Esmeli et al., 2021).
Interaction in online purchasing between firms and consumers happens on various platforms, for instance, the retailer's website and social media (Isa et al., 2016;Kusumawardani & Purniasari, 2021). In this interaction, there is supply and demand that results in the consumer purchasing a good or service online. In contrast to offline purchasing interaction (traditional interaction), online interaction can benefit consumers a lot. For example, it can save them time and money by allowing them to follow different product offers from home or anywhere (Al-Hattami, 2021; Kusumawardani & Purniasari, 2021). Consumers also save effort when purchasing online because they access all required information easily, read reviews, and compare alternative products (Fadhil et al., 2022).
Moreover, in a physical purchase interaction (offline), the sales representatives try to influence the buyers to buy the product. While in online purchasing, you are free to do as you will. This interaction has become increasingly important in the age of the pandemic (i.e., , where adherence to social distancing is important for human health and avoidance of infection with this most deadly pandemic (Al-Hattami 2021;Al-Qudah et al., 2022;Francioni et al., 2022;Nueangnong et al., 2020;Venkatesh et al., 2022). Yet, Yemen is different from other developing and developed countries in terms of facilities and technology infrastructure. In less developed countries (LDCs, like Yemen), survival is a crucial accomplishment, and this maximizes attention toward the role of digitalization in transforming the business process (Saleh & Manjunath, 2021). Due to tremendous challenges, LDCs have not adequately benefited from technological advancement, particularly Internet platforms (Isaac et al., 2018;Utoikamanu, 2019), which has led to creating significant gaps regarding the use of Internet platforms as well as causing a different willingness of consumers to interact in order to make online purchases.
Yemen, a LDC, 1 is characterized by slow growth in online purchases. This may be due to, among other things, inadequate facilities and infrastructure. This represents a big challenge for firms and can impact their development, mainly if they focus on Internet technology (Hanjaya et al., 2019). The online purchasing mechanism appears not to meet consumers' online purchase requirements. Consumer behaviour is a significant factor in facilitating online e-services. Thus, it is critical to recognize consumers' perceptions of these services and their intentions to use them (El-Ebiary et al., 2021). Analyzing consumers' perceptions of online purchasing is critical for e-commerce decision-makers to improve consumer expertise, provide recommendations, and boost revenue (Esmeli et al., 2021;Yang et al., 2022). Jain et al. (2022) also emphasized that research into consumer interaction with technology is one of the most critical areas of marketing.
Numerous studies have paid considerable attention to consumers' behaviour to understand their perceptions toward online purchasing in the context of mature (developed) e-commerce markets like China (Ying et al., 2021), Japan (Okamoto, 2021), India (Al-Hattami, 2021), Germany (Brusch & Rappel, 2020), Taiwan (Chen, 2019), Vietnam (Nguyen et al., 2019), Thailand (Driediger & Bhatiasevi, 2019), and Singapore (Hanjaya et al., 2019). To our knowledge, empirical evidence is still needed to investigate consumers' behaviour and understand their perceptions of online purchasing in the context of a less mature (less developed) e-commerce market such as Yemen. Consequently, further research is needed to understand how consumers of such markets compose their intentions to purchase online. Since cultural, economic, and technological differences between countries could cause different behaviour for consumers online, it becomes unsatisfactory to apply the reported results in mature (developed) countries directly to an international context (Çelik, 2011). Therefore, this paper explores factors that may impact consumer behaviour in the interaction intention to purchase online in Yemen as a less developed market.
Overall, a wide range of models seeks to understand the ambiguity regarding the use of Internet technology. Among these models, Davis's technology acceptance model (TAM) (Davis, 1989) is the most popular. TAM is a very effective model for predicting user interaction toward IT acceptance (Chen, 2019). It is widely accepted and validated by several studies as an accurate indicator of technology acceptance and use (Al Amin et al., 2021;Brusch & Rappel, 2020;Driediger & Bhatiasevi, 2019;Isaac et al., 2018;Ju et al., 2022;Nguyen et al., 2019;Okamoto, 2021;Peña-García et al., 2020;Sin et al., 2012). TAM implies that when introducing new technology to users, the users decide when and how to utilize this technology based on two main factors: perceived ease of use and perceived usefulness (Davis, 1989). Nevertheless, in emerging or less mature markets, the use or acceptance of technology can also be influenced by other factors such as technology infrastructure, technology anxiety, self-efficacy, and social influence (Al-Tuhaifi, 2017;Igbaria & Iivari, 1995;Venkatesh & Bala, 2008;Venkatesh et al., 2003). Although numerous papers have been conducted in the context of online purchasing based on the TAM theory, there is a paucity of research on how TAM and other variables (e.g., technology infrastructure and self-efficacy) affect interaction intention to online purchasing in LDCs, notably Yemen. Therefore, to cover this research gap, the current paper explores factors influencing consumer behaviour in the interaction intention to purchase online in Yemen, a less developed market. Specifically, the study's first aim is to establish the direct link between technology infrastructure, technology anxiety, self-efficacy, and social influence with interaction intention to purchase online.
As defined above, online purchasing is related to establishing and maintaining customer relationships by making full use of Internet platforms (such as websites and social media) to meet the needs of buyers (consumers). Such platforms have received great attention in developed countries, as the use of such platforms has become routine and important, especially in times of pandemics such as COVID-19 (Al-Hattami, 2021;Almaqtari et al., 2023;Nueangnong et al., 2020). Consumers in such countries do not face difficulties in using online platforms for shopping/purchasing due to the solid technology infrastructure in their countries (Asia Development Bank, 2018;OECD, 2021). This reflected positively on consumers' perceptions or acceptance of interaction with online purchasing platforms (OECD, 2021). Wigodsky (2004) indicates that technology infrastructure is the basis for the acceptance of technology-it is the medium used to connect all the parts to form a whole.
Yet, as one of the lowest-ranked countries, Yemen faces a challenge regarding developing technology infrastructure, whose costs cannot be ignored (Al-Halili & Hongxin, 2019). The Yemeni government has the sole monopoly on Internet services by taking Tele Yemen. From the perspective of Internet users, they are not satisfied with the lack of service and availability of Internet access (Al-Halili & Hongxin, 2019;Al-Nashmi & Amer, 2014). Therefore, technology infrastructure is a more proper factor in understanding the difference between developed and developing countries regarding the adoption of TAM. Hence, the second aim of the study is to investigate the moderating role of technology infrastructure in the relationship between determinants and interaction intention to purchase online.
The present study makes significant contributions to earlier literature. First, it presents a unique framework (based on TAM) to understand interaction intention to purchase online in a different context and culture. Hsu et al. (2017) confirmed that investigating various contexts and cultures is needed for shopping research. Granić and Marangunić (2019), in their systematic literature review, also stressed the significance of further validating TAM in various contexts and countries in order to promote its cross-cultural validity. Adapting a theory to different contexts is a popular method of expanding a theory; this assists in making the theory more powerful and increases its predictive validity (Williamson & Johanson, 2017). In this regard, the current study would expand TAM's scope and enhance its validity in a different context and culture. Second, the study expands TAM by the social factor, self-efficacy, technology anxiety, and technology infrastructure. This would promote TAM's interpretive power Mei & Aun, 2019;Peña-García et al., 2020).
The next section determines the literature review and hypothesis development. Section 3 shows the methodology. Section 4 highlights the analysis and results. Section 5 clarifies the discussion. Lastly, Section 6 outlines the conclusion.

Literature review and hypothesis development
The digital economy denotes a wide range of economic activities that utilize digital information and knowledge as key factors in business transactions. The economy is getting better and more efficient as digital technologies drive innovation, create jobs, and boost economic growth. The digital economy also permeates all aspects of society, influencing the way people interact and bringing about broad sociological changes (Asia Development Bank, 2018).
One of the developments of the digital economy that has permeated and dominated economies in both developed and developing countries is online purchasing platforms (OECD, 2021). These platforms have gained more popularity in developed countries due to the strong technological infrastructure in these countries, which increases the abundance of possibilities for online purchases (Asia Development Bank, 2018; OECD, 2021).
However, given the differences in the level of regulations and infrastructure, not all countries can take full advantage of the benefits offered by the digital economy (Almaqtari et al., 2023). Many times, people in LDCs do not have access to a basic online account, be it due to a lack of technology infrastructure, nationally accepted forms of identification, or socioeconomic barriers. To create more inclusion in the digital economy, there needs to be a deep understanding of the differences in access and adoption within the populations of different countries (Asia Development Bank, 2018). New issues related to social influence, technology anxiety, and self-efficacy also need to be addressed as LDCs' digital transformation intensifies (Al-Hattami, 2021aAl-Tuhaifi, 2017;Mei & Aun, 2019;Peña-García et al., 2020;Saprikis & Avlogiaris, 2021).
Adaptation to new technologies by consumers can be explained by a number of models such as IS success model, expectation-confirmation model, and TAM (Al-Hattami, 2021;Al-Hattami et al., 2021, 2022Al-Hattami, 2021b;Driediger & Bhatiasevi, 2019). TAM is one of the most commonly used theoretical models to explain why users accept or reject the technology and predict their behavior Granić & Marangunić, 2019;Peña-García et al., 2020).
Though the research on TAM offered insights into the use of technology, it focused on perceived ease of use and perceived usefulness as determinants of use/intent to use and ignored other external factors that could affect as determinants (Igbaria & Iivari, 1995). Perceived ease of use and perceived usefulness are only some of the appropriate factors determining technology acceptance, i.e., TAM needs to be extended by involving other factors (Xu et al., 2021). Original TAM constructs may not appropriately capture key beliefs affecting consumers' online purchasing behavior Peña-García et al., 2020). Therefore, in our context, other key factors should be considered to gain a better understanding of consumers' interaction intention to make purchases online. For example, technology infrastructure is a critical determinant of behavioural intention to use a certain technology, which drives the consumer to purchase online (Al-Tuhaifi, 2017; OECD, 2021). Lu and Yu-jen Su (2009) and Saprikis and Avlogiaris (2021) more specifically indicated that technology anxiety plays a significant role in consumers' interaction regarding online purchasing and needs further investigation. Among others, social influence and self-efficacy ought also to be considered to expand the behavioural intention to utilize online purchasing channels (Mei & Aun, 2019;Peña-García et al., 2020;Saprikis & Avlogiaris, 2021).
Overall, the current study expands TAM by incorporating the mentioned external factors affecting interaction intention to online purchasing and examining the model ( Figure 1) in a less developed country. The study also expands TAM by investigating "technology infrastructure" as a moderator. Shaikh et al. (2023) reported that using "moderators" could assist researchers in developing novel and exciting relationships between constructs in the online services field. Figure 2 presents the study model, while Table 1 contains a definition for each construct in the model.

Reference
Perceived usefulness (PU) PU refers to the extent to which a person thinks that the use of online purchasing channels will boost her or his job performance. (Venkatesh & Bala, 2008) Perceives ease of use (PEOU) PEOU is the degree to which a person thinks that the use of online purchasing channels will be possible and easy. (Venkatesh & Bala, 2008) Technology infrastructure (TI) TI refers to hardware components, software, network resources, and services required for the existence, operation, and management of online purchasing channels. (Techopedia, 2022) Technology anxiety (TA) TA refers to the tendency of an individual to be uneasy and/or worry about current or future use of online purchasing channels. (Igbaria & Iivari, 1995) Social influence (SI) SI refers to the change in consumer behavior toward online purchasing because of someone, intentionally or unintentionally.

(Mei & Aun, 2019)
Self-efficacy (SE) SE refers to the degree to which an individual believes that he or she has the capability to use online purchasing channels. (Venkatesh & Bala, 2008) Interaction intention to purchase online (IIPO) IIPO is a consumer's willingness to purchase a certain product or service online. (Driediger & Bhatiasevi, 2019) Based on the fact that if consumers find online purchasing a helpful channel, they will feel more determined to use it, prior research revealed that PU positively impacts behaviour intention (Al-Hattami, 2021;Brusch & Rappel, 2020;Driediger & Bhatiasevi, 2019;Rehman et al., 2019;Sin et al., 2012). Suppose the consumers realize that online purchasing requires no effort. In that case, the motivation to purchase is more likely as it is known that purchasing involves investing in terms of time, money, and mobility (Peña-García et al., 2020;Srivastava & Thaichon, 2023). Accordingly, consumers' expectations for the performance of online purchasing channels affect their interaction with buying. Thus, improving the efficiency of online purchasing channels would make consumers more inclined to utilize them. Such an argument is not empirically proved in Yemen, as a less developed country. Therefore, the first hypothesis is framed as follows:

Perceives ease of use (PEOU)
Internet sites are information stores that help consumers search for information. The ease of use of an Internet site is critical, as it is a vital motivator for consumers who select to use a certain purchasing site (Chaudary et al., 2014). According to Driediger and Bhatiasevi (2019), ease of use, directly and indirectly, impacts customers' inclination to shop online. The spillover impact on intention occurs through perceived usefulness, as the easier a technology is to use, the more valuable it is. Other studies have found evidence that interaction intention toward online purchasing positively relates to ease of use (Al Amin et al., 2021;Hanjaya et al., 2019;Sin et al., 2012). However, evidence of this relationship in the context of Yemen is still lacking. Thus, it is assumed that: Venkatesh et al. (2003) stated it was important to look at anxiety as a negative emotional response of online users. Research on online shopping has shown (Lu & Yu-jen Su, 2009;Saprikis & Avlogiaris, 2021) that anxiety has a significant effect on the likelihood of buying online. In addition, Dewi et al. (2020) reported that anxiety is a significant and practical obstacle to consumers' responses to Internet purchasing. Venkatesh et al. (2003) noted that the worry is mostly about the dangers of giving out personal or banking information when making an online purchase.

Technology anxiety (TA)
Notably, anxiety as a determinant of online purchasing showed a negative effect on behavioural interaction intention to purchase (Dewi et al., 2020;Faqih, 2016;Lu & Yu-jen Su, 2009;Saprikis & Avlogiaris, 2021). However, detection of such an impact in the context of LDCs such as Yemen is still limited. Thus, it is proposed that:

Social influence (SI)
The connection between social influence and behavioural intention is vastly debated (Rahi et al., 2019). The literature identifies social influence as an essential factor in consumer behaviour since no person in this universe can completely escape the impact of others (Mei & Aun, 2019). The degree to which others interact with online purchasing channels can be influenced by the positive experiences of members of society, such as friends and relatives (Brusch & Rappel, 2020;Mensah, 2020). Dewi et al. (2020) and Al Amin et al. (2021) concluded that social influence positively impacts interaction intention to purchase online. In the current study's context, it can also be proposed that:

Self-efficacy (SE)
As self-efficiency reflects one's confidence in implementing behaviour, studies on technology adoption have extensively used this concept (Gupta & Arora, 2017;Shukla et al., 2021). In our context, the consumer's ability to access a personal computer and the Internet and know how to make an online purchase are key factors influencing the consumer's interaction intention of making an online purchase (Boyle & Ruppel, 2006;Peña-García et al., 2020). Many studies concluded that self-efficiency positively impacts online purchasing behaviour (Mensah, 2020;Peña-García et al., 2020). Therefore, in our context, self-efficiency is likely to influence beliefs and behaviour toward online purchasing. That is, the sixth hypothesis is: H6. Self-efficacy positively impacts intention to purchase online

Technology infrastructure (TI)
LDCs like Yemen must capitalize on technology for development and growth (Al-Hattami et al., 2022;Al-Hattami, 2022 -Tuhaifi, 2017). Policies to improve access to infrastructure and connectivity geared towards improving digital skills and spreading the use of the Internet, can accelerate the spread of online platforms and promote countries' economic wealth (OECD, 2021). Therefore, improving the infrastructure concerning online purchasing channels would make consumers more interactive in utilizing these channels. Based on that, this study assumes that:

H7. Technology infrastructure positively impacts interaction intention to purchase online
Unlike developed countries, LDCs (including Yemen), face a challenge regarding developing technology infrastructure (Al-Halili & Hongxin, 2019;OECD, 2021). A robust infrastructure of technology is the key to the acceptance of any technology, especially frontier technologies such as online purchasing platforms (OECD, 2021;UNCTAD, 2021). The Report of UNCTAD (2021) argued that frontier technologies are necessary for sustainable development and contribute to increasing equality. The least developed nations and low-income emerging nations cannot afford to miss the novel wave of fast technological progress. These countries will need to promote the usage, acceptance, and adaptation of frontier technologies in order to capitalize on this novel technological revolution. A balanced approach to building a strong technology infrastructure and promoting frontier technologies is essential for success in the twenty-first century.
Therefore, technology infrastructure is a more important factor in understanding the difference between developed and developing countries regarding the interaction intention to purchase online. Hence, the current study investigates the moderating role of technology infrastructure in the relationship between determinants and interaction intention to purchase online. Some studies suggest that using moderators would enhance behavioural intent (Almaqtari et al., 2023;Rehman et al., 2019). Since no study has explored the technology infrastructure as a moderator of interaction intention to purchase online in LDCs, this study assumes:

Methodology
The study focuses on identifying the key factors determining consumer behaviour toward the interaction intention to purchase online in Yemen as a less developed market. For that, the quantitative approach is applied. The study targeted certain categories, including lecturers and students at universities recognized by the Ministry of Higher Education (MHE) in Yemen. Only educated people were considered because they know about online purchasing platforms more than uneducated people. The data for the study was collected through a stratified sampling method. This method is the most common among probability methods (Reis et al., 2018). Moreover, this method offers a representative sample (Al-Hattami, 2022; Ros & Guillaume, 2019).
An online questionnaire was developed based on the previous studies to gather the data from respondents, as the questionnaire is believed to be suitable for examining the study hypotheses. The online questionnaire is considered the most suitable tool for gathering data, particularly during the COVID-19 pandemic, where social distancing is advisable (Al-Hattami, 2021;Francioni et al., 2022). It is also considered an inexpensive and effective technique for data collection, particularly with a substantial sample size (Thomas, 2004). The questionnaire was divided as follows: following the personal information, the questions related to the study model were presented, namely, perceived usefulness, perceived ease of use, technology infrastructure, technology anxiety, social influence, self-efficacy, and interaction intention to purchase online. Furthermore, the study applied the five-Likert scale for collecting data using the so-called Google Doc.
The questionnaire link has been sent to respondents from June-July 2022. Accordingly, the total number of received responses was 277, of which 273 were suitable and had enough statistical power to conduct the needed analysis (Hair et al., 2021); this number was considered the final sample for the present study. The breakup of final sample comprised 61.5% male and 38.5% female; most of them (58.6%) with an age range of 25-35 years. Most of them (83.2%) were Master's or Ph.D holders, and the rest were bachelor's degree students.
The partial least squares structural equation modeling (PLS-SEM) was used for data analysis and testing of hypotheses. It is an adequate technique with no assumptions about data distribution and variables with adequate prediction. The PLS-SEM is employed extensively in marketing, management, and IS research (Dash & Paul, 2021;Hair et al., 2021;Urbach & Ahlemann, 2010). Moreover, many researchers consider it a preferred technique for measuring and analyzing the moderator (Rehman et al., 2019).

Analysis and results
In the first step of PLS-SEM, the measurement model is looked at in terms of its reliability and validity. If all the measurement model's conditions are met, the next step is to evaluate the structural model by testing hypotheses (Almaqtari et al., 2023;Hair et al., 2021). Yet, issues of multicollinearity and common method bias (CMB) should be examined before proceeding to the analysis. The multicollinearity issue, which can be determined by the variance inflation factor (VIF), is not welcome in research. For CMB, the literature indicated the importance of examining its effect on analysis findings (Jordan & Troth, 2020). Based on Kock (2015), VIF >3.3 reports the existence of the multicollinearity issue and that a model could be a CMB. As in Table 3, the highest VIF score is 2.423 (<3.3), indicating no multicollinearity or CMB issue in this research. Table 2 shows the measurement model's conditions. When comparing these conditions with what is achieved in the current study, it is clear that all the conditions of the measurement model are met (see Table 3).
The links between variables (hypotheses) are expressed by path coefficients (β), as clarified in Figure 2. Scores of t and p are utilized to assess whether β-values are significant or not. A t-value of >1.96 and a p-value of <0.05 are recommended (Hair et al., 2021). Based on that, all the study hypotheses (except H1, H3, H4, H8, and H11) were supported (Table 4). R 2 explains the variance in the dependent variable resulting from independent ones. According to Hair et al. (2021), the R 2 level judgment is based on the study field identified; e.g., an R 2 score of 0.20 is considered significant in areas like consumer behaviour. The interpreted variances (R 2 ) of PU and IIPO are 0.517 and 0.705, respectively ( Figure 2). This asserts that the proposed model substantially interprets the dependent variables.

Discussion
In this empirical research, the Yemeni consumer's interaction intention to purchase online was analysed as a context for LDCs. The research developed an extended model of TAM by adding technology infrastructure (direct/moderate), technology anxiety, social influence, and self-efficacy as independent variables that impact IIPO. After the bootstrapping procedure with 5000 subsamples, the findings significantly supported all hypotheses except H1, H3, H4, H8, and H11. The study did not support perceived usefulness as a significant determinant of IIPO (H1, β = 0.071, p > 0.05). This contradicts the TAM theory and previous research (Al-Hattami, 2021;Brusch & Rappel, 2020;Davis, 1989;Rehman et al., 2019). However, this result is in line with previous research by Nguyen et al. (2019), who confirmed that PU has no effect on intention toward online purchasing in an emerging economy.
PEOU was found to be a key driver of PU (β = 0.719, p < 0.05). That is, H2 is valid. According to this result, PU is strongly determined by how easy it is to order/purchase a product or service through online purchasing channels. This result introduces empirical support for the TAM and also supports the past research by Faqih (2016), Brusch and Rappel (2020), Nguyen et al. (2019), and Yuen et al. (2022). Consumers believe that improving PEOU would increase their perceptions of the usefulness of online purchasing by lowering their physical and mental effort. Those consumers want shopping  (Hair et al., 2021).
Yes Table 3 -Cross loadings (CL): The CL of each variable must be above CL of all other variables in the variable's row and column (Urbach & Ahlemann, 2010).

Yes
Appendix 2

Validity (discriminant)
Fornell-Larcker standard: the correlation values of each variable have to be lower than √AVE (Hair et al., 2021).
Yes Table 3 Al -Hattami et al., Cogent Social Sciences (2023) to be easy and possible (Venkatesh & Bala, 2008). Shekhar and Jaidev (2020) argued that technologies seen as easy to use would indirectly motivate consumers to buy online. The effort put out can be used to determine how simple a given technology is to use, i.e., making a lower effort means higher value for the user and, therefore, a positive attitude toward the technology (Ju et al., 2022).
PEOU is a closely related aspect of the consumers' intention to make an online purchase in Yemen, where online purchasing is still a novel activity. Yemeni consumers may feel uneasy engaging in this novel shopping scenario because they lack sufficient knowledge and skills to carry out online purchase tasks; thus, virtual storefronts should be simple enough for them to use and explore what fits them entirely. Online stores in Yemen should strive to design websites that are easy to interact with and use. This would reduce consumers' efforts related to purchasing products online.
PEOU is considered one of TAM's key factors, but it has not been asserted in the current paper as an important factor directly affecting IIPO (H3, β = −0.003, p > 0.05). Although this result contradicts numerous prior studies (Al Amin et al., 2021;Francioni et al., 2022;Moslehpour et al., 2018;Rehman et al., 2019), it is in line with other studies (Brusch & Rappel, 2020;Okamoto, 2021). This proves that the ease of using online purchasing channels may not be an important factor in the interaction intention of online purchasing. A possible reason to interpret this finding is that respondents have a university education background, hence using the Internet for purchasing may be easy for them compared to those with a low educational background; this was also confirmed prior by Hing and Vui (2021). Anxiety (β = 0.009, p > 0.05) has not been shown to have a negative impact on interaction intention to purchase online. Therefore, H4 is invalid. This result reports that anxiety might not prevent people from purchasing online. This result contradicts many studies, such as those by Saprikis and Avlogiaris (2021), Dewi et al. (2020), andFaqih (2016), who found anxiety to impact IIPO negatively. To some extent, this result can be justified by the fact that most study participants were postgraduates (intellectuals) and, therefore, may not see anxiety as a negative factor. Çelik (2011) also reports that consumers with less knowledge of online purchasing technology may have greater anxiety about their use and vice versa.
For social influence, it was found to positively impact IIPO (H5, β = 0.369, p < 0.05). This is interpreted by the fact that the more people (relatives, friends, and other people in the community) who think using online purchasing channels is a good idea, the more consumers intend to use such channels. This result is supported by past research (Alkhwaldi et al., 2023;Dewi et al., 2020;Dwivedi et al., 2019;Faqih, 2016;Rehman et al., 2019;Sin et al., 2012). Miranda et al. (2014) also proved that the viewpoints of friends and relatives exert a positive impact on consumers' behaviour toward IIPO. Yemen is a collective community where the needs and goals of the group as a whole are over the needs and desires of each individual. In such communities, consumers will be more open to social pressure while making online purchasing decisions (Hing & Vui, 2021). Thus, to encourage interaction toward online purchasing, Yemeni online retailers should spread and promote the culture of online purchasing among community members.
The study concluded that IIPO is positively impacted by self-efficiency (H6, β = 0.169, p < 0.05). A similar outcome was observed in the previous research (Boyle & Ruppel, 2006;Mensah, 2020;Peña-García et al., 2020;Shukla et al., 2021). The present study documents the direct impact of self-efficiency on the behavioural intention to interact with online purchasing channels in the context of Yemen. This means that increasing consumers' self-efficiency increases their intention to use online purchasing channels. One of the implications of this result is that Yemeni retailers should offer potential consumers training or Besides R 2 , the predictive relevance (Q 2 ) is presented in Table 4 and was found to be satisfactory, i.e., Q 2 > zero (Cohen, 1988). Q 2 > zero displays the predictive relevance of the model. Finally, the model fit is denoted in two ways: the SRMR and the overall model fit standard. The SRMR was discovered to be 0.07, which was less than the 0.08 criterion (Hair et al., 2021). The overall model fit is measured by the goodness-of-fit (GoF) index, which can be computed as GoF = √ (AVE average * R 2 average) (Tenenhaus et al., 2005). A GoF > 0.25 denotes a high overall model fit. As apparent in Appendix 3, GoF was 0.660, denoting a high overall model fit.
educational programs on online purchasing channels. This approach would further improve their selfefficiency by clarifying how to buy products/services online.
Lastly, technology infrastructure (β = 0.174, p < 0.05) has positively impacted interaction intention to purchase online. Therefore, H7 is valid. This result supports Al-Tuhaifi (2017) and Al-Nashmi and Amer (2014). This shows that if the technology infrastructure is strong, online purchasing channels will be strengthened and accepted. The respondents believe that the contribution of technology is reflected in rapid and inexpensive access to the Internet and reliable and continuous energy sources (e.g., electricity). Dwivedi et al. (2019) noted that the technology infrastructure concerning technology services, such as help desks and training programs, could also strengthen the user's behavioural intention toward the technology. Therefore, the technology infrastructure should be considered to motivate the acceptance and interaction of online purchasing channels among Yemeni consumers.
Concerning the effect of technology infrastructure as a moderator, the results support H9, H10, and H12, but do not support H8 and H11. Specifically, technology infrastructure moderated the relationship between PEOU and IIPO (H9, β = 0.204, p < 0.05), AT and IIPO (H10, β = −0.119, p < 0.05), although a direct relationship was found to be insignificant (see H3 and H4 in Table 4), which is the novel contribution of this study in this context. This novel contribution comes since no study has explored the technology infrastructure as a moderator of IIPO in LDCs. This would prompt future research to further research and investigate this variable. Technology infrastructure also moderated the association of SE and IIPO (H12, β = 0.218, p < 0.05). Overall, businesses with higher levels of technology infrastructure could promote the association between (PEOU, AT, SE) and IIPO. The results also showed that technology infrastructure has no moderate impact on PU-IIPO and SI-IIPO relationships; this suggests that levels of technology infrastructure cannot be estimated by the interactions of customer PU and SI with IIPO, respectively.

Conclusion
User acceptance of new technology services is often seen as one of the most important areas of research in modern information systems literature (Dwivedi et al., 2019). Purchasing products online is a new trend among consumers in LDCs such as Yemen. Therefore, it is necessary to comprehend the determinants of consumer intention toward its acceptance and use. So far, there is a scarcity of empirical research investigating online purchasing behaviour in LDCs. In Yemen, the factors influencing IIPO have not been empirically determined. Thus, this research would contribute significantly to theory as it was conducted in a limited research context and culture (non-Western). Furthermore, the paper focuses on the moderating role of technology infrastructure. Prior research (e.g., Nguyen et al., 2019) did not contemplate the moderating role of technology infrastructure in the relationships of the variables with IIPO. Therefore, the present research would assist in deepening our knowledge of this dimension as it adds value to the current literature by illustrating the moderate impact of technology infrastructure on the relationship between (PU, PEOU, TA, SI, SE) and IIPO. Such an impact has not been previously investigated.
This study determined the key factors to consider for predicting the behavioural intentions of Yemeni consumers to interact with online purchases. Additionally, it developed a model that would assist online retailers in understanding the determinants of consumers' intentions to purchase online. Exploring these determinants, especially in this period while the online retail market in Yemen is still developing, is crucial for stakeholders to ensure the success of this emerging market. This would help stakeholders clarify online retail strategies for website design, online advertising (e.g., via social media), market segmentation, sales promotion, and product diversity (El-Ebiary et al., 2021;Hanjaya et al., 2019;Mukherjee & Banerjee, 2019;Rehman et al., 2019). As such, the adoption and interaction of online purchasing will result in the existence and use of various products among consumers, ultimately raising their standard of living and driving up the GDP (Usman & Kumar, 2021). This study has some limitations that should be pointed out so that they could be fixed in future research. First, the present research investigated the direct impact of PU, PEU, TI, TA, SI, and SE on IIPO. In future research, there is a need to consider additional variables such as trust. Second, this study has yet to evaluate the impact of the demographic variables. Further research may be conducted to know the role of such variables (e.g., a particular gender, age, or educational level) in the acceptance of online purchasing. Third, besides the current quantitative approach taken in this paper, it is suggested to use a qualitative approach in future studies, which could lead to broader results. Fourth, the results were based on 273 responses, and future research can increase the sample size. Sixth, the sample did not consider people without a university education, who could buy online. Additionally, not all people with a university education buy online. This would limit the generalization of the results. Lastly, while we limited ourselves to a certain culture, the research could be expanded to include a multicultural comparison.
PU2: Using online purchasing channels enables one to browse and purchase products/services easily.
PU3: Using online purchasing would enable me to order products/services from home or anywhere.
PU4: Using online purchasing could overcome the difficulties for elderly or physically disabled people.

PEOU (Venkatesh and Bala, 2008)
PEOU1: The online purchasing process is easy to learn. PEOU2: The online purchasing process would not require much effort.
PEOU3: Overall, I find online purchasing possible and easy.

TA (Venkatesh et al., 2003)
TA1: Using the Internet in purchasing process worries me.
TA2: I feel worried to think that I may lose a lot of information by clicking on the wrong key.

SE (Boyle and Ruppel, 2006; Peña-García et al., 2020)
If online purchasing channels are widely available in my country: SE1: I am confident that I could use these channels.
SE2: I could easily use the Internet to find information about a product and purchase it. SE3: I expect to become very skillful in shopping using online purchasing channels.

TI (the authors)
TI1: An online purchase requires fast and inexpensive internet access.
TI2: Reliable and continuous power sources (electricity) are also required to make an online purchase and follow up the order.
TI3: Overall, sufficient technology infrastructure is required to make an online purchase.
IIPO2: I intend to use online purchasing when the service becomes widely available.
IIPO3: If there are good and secure online purchasing channels, I will recommend them to others.